The strategy for the 2000 Update is to revisit the tropical FRA 1990 sampling units, and update them with a third date of imagery for the reference year 2000. Such a survey will provide up-to-date information on change processes at pan-tropical, regional and ecological level, as well as new and unprecedented insights on deforestation trends over the periods 1980-1990 and 1990-2000.
The new information on deforestation trends will be used to improve the efficiency of the future pan-tropical sample of the World Forest Survey and will represent a unique reference for spot checks and deforestation modeling of existing information.
The 2000 Update is currently implemented with the financial support of the Government of Finland (Project GCP/INT/723/FIN), and of FAO Regular Programme.
The main activities of this project include the refinement of the methodology and the re-visit of the FRA 1990 pan-tropical sampling units by adding recently acquired high-resolution data. Along with these technical issues efforts will be made to promote the harmonization of ongoing regional and global programmes and the institutionalization of resource monitoring as an important ingredient to planning and policy formulation at national and international level.
The output of this project will include reliable updated estimates of forest cover state and change, including current trends, and detailed descriptions of the change processes affecting the tropical forest resources by geographical (global, regional) and ecological reporting units. The new data set will complete the FRA 1990 multi-purpose database with recent spatial and statistical data, which can serve to a) improve the deforestation model (through FORIS)2 and b) provide baseline data for environment oriented studies.
The methodology followed for the 2000 Update is strongly linked to the one followed for FRA 1990. In view of the paramount importance of methodological consistency within time series for a reliable estimation of changes, the analysis of the recent "2000" images must be done coherently to the format of the pre-existing 1980 and 1990 data sets.
This methodological approach presents practical and conceptual aspects.
It is practical to continue with visual interpretation of analogue data since that was the approach followed in the past. At this point, changing from the analogue to the digital approach would mean purchasing and processing of digital data for all historical images and repeating all analyses.
It is conceptually essential to continue with the interdependent interpretation approach developed and followed during FRA 1990 since this is the only one that ensures high consistency in the estimation of changes through the elimination of error propagation. The interdependent interpretation procedure is now applied to the three dates using the intermediate 1990 image (T2) as "pivot". It’s not the simple interpretation of one additional image but rather the addition of new evidence into a three-date reference system. The contribution of the third date (T3) in this case is not only used for the estimation of the changes between T2 and T3 but also to improve the reliability and consistency of the entire T1-T2-T3 series. In fact, in addition to providing evidence of the changes recently occurred, T3 often contributes to a better understanding of the terrain condition, vegetation types and land-uses. This new knowledge is therefore inserted into T2 and T3 interpretations, which are revised to improve the assessment of state as well as change. The consistency within the entire time series is extremely important if we want to achieve the challenging objective of reliably measuring not only recent change rates but overall trends.
1.2 Expected 2000 Update Results
As example of an individual sampling unit of the 2000 Update study, Annex 1 shows the images and the results produced for a location in Mato Grosso, Brazil. The statistical results reported include the sequential transition matrices for the periods 1972-1989 and 1989-1997, the periodic and annual deforestation rates and the deforestation trend estimated as the percent variation of annual rates. The information contained in the transition matrices is also represented in form of biomass flux diagrams, which help to visualize and compare the processes of change observed over the two periods. In addition to clearly displaying the quantity of change, the analysis of sequential diagrams indicate if the process have changed over time highlighting, perhaps, changes in the cause-effect mechanisms.
The results aggregated at pan-tropical, regional and ecological level will represent the main contribution of the survey. The reporting units will be the same of Forestry Paper 130, which reports the results of the 1990 survey. The thematic content, however will be new in view of the new period observed and the unprecedented analysis of trends.
As preliminary example of trend analysis, the graph in Figure 2 shows the scattergram of pre- and post-90 annual deforestation rates observed on a subset of the South America sample. The subset here represented is composed by 12 sampling units that were recently re-visited in the framework of a cooperative FAO FRA/JRC TREES activity3.
Figure 2: Analysis of deforestation trend based on a sub-set of the South America sample. Deforestation rates refer to F3 definition of forest. Input data is shown in Annex 2.
Figure 3: Biomass flux diagram 1980 – 1990. Based on FRA 1990 standardized transition matrices of the re-visited sampling units
Figure 4: Biomass flux diagram 1990 – 1996. Based on aggregation of transition matrices of the re-visited sampling units
The results displayed in Figures 2, 3 and 4 are not statistically representative of South America, since they are based on a small number of units. Nonetheless, the results reveal interesting new information on the distribution of current deforestation and its trend. In the graph in Figure 2 the trend is estimated as the percent ratio of the annual deforestation rate of the second period over that of the first period (input data is tabulated in Annex 2). The distribution of data points shows that there is a wide variation between the deforestation rates observed in the two periods. It is quite evident however, that the negative trend (increment of deforestation) is mainly shown by the units which already had a high deforestation rate. From this limited number of sampling units it appears that the areas with low pre-’90 deforestation rates present an even lower rate after 1990, while the areas with high pre-’90 deforestation present an even higher post-’90 rate. The trend appears clearly negative, with a coefficient of 1.52, which means some 52 percent increase in annual deforestation rate. There is also a very clear geographic distribution of rates and trends. The sampling units located in the Brasilian States of Mato grosso and Rondonia show the highest deforestation rates and a consistent negative trend.
In addition to the quantitative estimation of deforestation trends, (change of deforestation rates) it is interesting to analyze the variations occurring to the change processes before and after 1990, which we may call process trends. This analysis helps to understand how the causes and mechanisms associated to deforestation, degradation, etc. are changing in time.
In order to compare the change processes of the two periods, (which are based on different lengths of time), the analysis is based on the percentages of each period’s change.
A comparison of the two diagrams indicates that the dominance of the transition from closed forest to other land cover, for the 1990-1996 study period, is greater than that of the 1980-1990 period. Expansion of large-scale cattle ranching and, to a lesser degree, permanent agriculture, is becoming the dominant cause of deforestation for these study sites. Subsistence farming associated to re-settlement programmes, indicated by the transition closed forest to short fallow, that represented a significant share of the total change in the pre-’90 period, is now almost absent. This could be the effect of changes in Brazilian policies regarding resettlement programs.
2FORIS (Forest Resources Information System) is a statistical database which stores national/sub-national forest resources data (forest area, functions, growing stock, volume) developed under FAO's FRA 1990 project.
3 Project GCP/INT/131/EC “Estimation of Recent Deforestation Rate in South America”
